Proton Sensors in the Pore Domain of the Cardiac Voltage-gated Sodium Channel*

Background: Protons modify cardiac sodium channel function, potentially contributing to cardiac arrhythmia during and following ischemia. Results: Protons binding amino acids at the pore alter cardiac sodium channel function. Conclusion: Sodium channel pore protonation mediates proton block and destabilization of sodium channel slow inactivation. Significance: Understanding sodium channel proton modulation is necessary to understand the pathophysiology of cardiac acidosis. Protons impart isoform-specific modulation of inactivation in neuronal, skeletal muscle, and cardiac voltage-gated sodium (NaV) channels. Although the structural basis of proton block in NaV channels has been well described, the amino acid residues responsible for the changes in NaV kinetics during extracellular acidosis are as yet unknown. We expressed wild-type (WT) and two pore mutant constructs (H880Q and C373F) of the human cardiac NaV channel, NaV1.5, in Xenopus oocytes. C373F and H880Q both attenuated proton block, abolished proton modulation of use-dependent inactivation, and altered pH modulation of the steady-state and kinetic parameters of slow inactivation. Additionally, C373F significantly reduced the maximum probability of use-dependent inactivation and slow inactivation, relative to WT. H880Q also significantly reduced the maximum probability of slow inactivation and shifted the voltage dependence of activation and fast inactivation to more positive potentials, relative to WT. These data suggest that Cys-373 and His-880 in NaV1.5 are proton sensors for use-dependent and slow inactivation and have implications in isoform-specific modulation of NaV channels.

Protons impart isoform-specific modulation of inactivation in neuronal, skeletal muscle, and cardiac voltage-gated sodium (Na V ) channels. Although the structural basis of proton block in Na V channels has been well described, the amino acid residues responsible for the changes in Na V kinetics during extracellular acidosis are as yet unknown. We expressed wild-type (WT) and two pore mutant constructs (H880Q and C373F) of the human cardiac Na V channel, Na V 1.5, in Xenopus oocytes. C373F and H880Q both attenuated proton block, abolished proton modulation of use-dependent inactivation, and altered pH modulation of the steady-state and kinetic parameters of slow inactivation. Additionally, C373F significantly reduced the maximum probability of use-dependent inactivation and slow inactivation, relative to WT. H880Q also significantly reduced the maximum probability of slow inactivation and shifted the voltage dependence of activation and fast inactivation to more positive potentials, relative to WT. These data suggest that Cys-373 and His-880 in Na V 1.5 are proton sensors for use-dependent and slow inactivation and have implications in isoform-specific modulation of Na V channels.
Voltage-gated sodium (Na V ) 4 channels are responsible for action potential propagation in most excitable cells. Each channel is composed of a pore-forming ␣-subunit and one or more modulating ␤-subunits (1,2). The ␣-subunit forms the functional pore of the channel and is composed of four homologous domains (DI-DIV) consisting of six, transmembrane helices (S1-S6) (1,3). Helices S1-S4 form the channel's voltage-sensing domain, primarily mediated by a high concentration of positively charged arginines and lysines positioned at every third residue within each S4 helix (4). The S5 and S6 helices and the extracellular loops linking them (p-loops) combine to form the functional pore and selectivity filter of the channel (5).
Two processes regulate Na V channel availability: fast inactivation (FI) and slow inactivation (SI). FI is believed to occur through occlusion of the intracellular mouth of the pore by the intracellular DIII-DIV linker (6). This process is called FI because it occurs in the time frame of several milliseconds. In contrast, Na V channels undergo slow inactivation in response to prolonged (Ͼ0.5-s) depolarization or rapid, successive depolarizations. This process is known as SI because its onset and recovery occur in the time frame of seconds to tens of seconds (7). SI is also molecularly and pharmacologically distinct from FI (7). Intermediate, slow, and ultraslow inactivated states have been described and occur in the 0.5-1 s, 1-60 s, and Ͼ60 s time domains, respectively (8,9). The presence of sequential inactivated states, dependent on the duration and frequency of depolarization and distinct from FI, was suggested to be a continuum of slow inactivated states (7).
The mechanism of SI is poorly understood but is thought to be a collapse of the pore mediated through long distance interactions within the channel. Several mutations at the external pore disrupt SI (8 -10). Metal cations at the external pore inhibit SI (11), and structural rearrangement at the external pore occurs concurrently with SI onset (12,13). Further, the probability that a channel will slow inactivate varies between isoforms (ϳ50% in Na V 1.5 and ϳ90% in Na V 1.4), a difference that is mediated by p-loop residues of the DII S5-S6 linker (8,14). C-type inactivation, the potassium channel equivalent of Na V channel SI, involves constriction at the selectivity filter coupled with the movement of the activation gate (15,16). Most recently, several crystal structures of bacterial Na V channels suggest that collapse of the channel's selectivity filter contributes to slow inactivation (17,18).
Additionally, mutagenic replacement of residues Ile-1303, Phe-1304, and Met-1305 (Na V 1.4 numbering) of the DIII-DIV linker with QQQ abolishes FI but increases the probability for SI (19). Mutating the negatively charged residues adjacent to the IFM sequence alters the properties of SI (20). Further, mutations involving the cytoplasmic S4-S5 linkers as well as the S4, S5, and S6 helices have been shown to affect Na V channel SI (21)(22)(23)(24)(25)(26)(27)(28). Capes et al. (29) showed that prolonged depolarizations that induced slow inactivation inhibited gating pore currents through the DIV voltage-sensing domain and demonstrated that movements within the Na V channel voltagesensing domain and selectivity filter are coupled.
Protons destabilize the fast and slow inactivated states of Na V 1.5 channels, the primary Na V channel isoform found in cardiac tissue. During cardiac ischemia, extracellular pH drops from pH 7.4 to as low as pH 6.0 within ϳ10 min of its onset (30). Changes in Na V 1.5 channel function during cardiac ischemia are believed to contribute to cardiac arrhythmia (31). Protonation of p-loop amino acids causes a reduction of single channel conductance, in skeletal muscle Na V 1.4 channels, with a pK a of pH ϳ6.0 and complete block at pH ϳ4.0 (32,33). The amino acids that may underlie changes in Na V 1.5 channel gating during extracellular acidosis are unknown.
The goal of this study was to identify proton-sensitive residues that mediate the changes in Na V 1.5 channel kinetics during extracellular acidosis. We expressed wild-type and mutant constructs of the human variant of the cardiac Na V channel, hNa V 1.5, in Xenopus oocytes. We demonstrate that p-loop residues mediate proton modulation and block of Na V 1.5 channels and contribute to Na V 1.5 channel SI stability.

MATERIALS AND METHODS
Molecular Biology-The human variant of the pore-forming ␣-subunit, hNa V 1.5, in SP64T was graciously donated by Dr. Chris Ahern (University of British Columbia). hNa V 1.5 and H880Q DNA was linearized using XbaI (Invitrogen). Transcription was completed using a Sp6 mMESSAGE mMA-CHINE high yield capped RNA transcription kit (Applied Biosystems, Carlsbad, CA). The C373F mutant was graciously donated by Dr. Mohamed Chahine (Laval University, Canada). C373F DNA was linearized using NotI (New England BioLabs, Pickering, Canada). C373F transcription was completed using a T7 mMESSAGE mMACHINE high yield capped RNA transcription kit (Applied Biosystems). The H880Q mutant construct was generated using the QuikChange method (Stratagene, Mississauga, Canada) with primers synthesized by Sigma Genosys (Oakville, Canada). All constructs were sequenced with the use of Eurofins MWG Operon (Huntsville, AL).
Oocyte Preparation-Oocyte preparation was completed as described previously (34). Briefly, female Xenopus laevis were terminally anesthetized using tricaine solution (2 g/liter). Oocytes were surgically removed, and theca and follicular layers were enzymatically removed by ϳ1-h agitation of semi-intact lobes in a calcium-free solution, containing 96 mM NaCl, 2 mM KCl, 20 mM MgCl 2 , 5 mM HEPES and supplemented with 1 mg/ml type 1a collagenase. Collagenased oocytes were then washed and sorted in calcium-free solution. Stage V-VI oocytes were injected with 50 nl of cRNA encoding either WT or mutant hNa V 1.5. Injected oocytes were incubated at 19°C in SOSϩ medium containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, 2.5 mM sodium pyruvate, supplemented with 100 mg/liter gentamicin sulfate and 5% horse serum for 3-10 days prior to recording. All surgical and animal care procedures were completed in accordance with the policies and procedures of the Simon Fraser University Animal Care Committee and the Canadian Council of Animal Care (35).
Data Acquisition-Data were acquired as described previously (34). Briefly, recordings were made using a CA-1B amplifier in the cut-open mode. Data were low pass-filtered at 10 kHz, digitized at 50 kHz, and then recorded using Patchmaster (HEKA Electronics, Mahone Bay, Canada) running on an iMac. Cells were permeabilized via bottom bath perfusion with intracellular solution containing 9.6 mM NaCl, 88 mM KCl, 5 mM HEPES, 11 mM EGTA, titrated to pH 7.4, and supplemented with 0.1% saponin. After 1-2-min exposure, saponin-free intracellular solution was washed in. Extracellular solution of the top and middle chambers contained 96 mM NaCl, 4 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES. For recordings completed below extracellular pH 6.5, HEPES was substituted with 5 mM MES. Bath chambers were temperature-controlled at 22°C using a Peltier device run by a TC-10 temperature controller (Dagan, Minneapolis, MN).
Pulse Protocols-Cells were maintained at a holding potential of Ϫ100 mV between all protocols. Unless otherwise stated, leak subtraction was completed online, using a Ϫp/4 protocol from a holding potential of Ϫ100 mV. Current/voltage relationships, steady-state FI (SSFI), and FI recovery and onset were measured as described previously (34).
Prolonged versions of the FI protocols were used to measure SI. Steady-state SI (SSSI) was measured by holding the membrane potential at Ϫ150 mV for 30 s to recover channels from the slow inactivated state and then stepping to 60-s conditioning pulses between Ϫ150 and Ϫ10 mV in 20-mV increments. The membrane potential was then hyperpolarized to Ϫ150 mV for 20 ms to allow recovery of fast inactivated but not slow inactivated channels and then depolarized to a 0-mV test pulse to measure the remaining channel availability. Several cells were administered a Ϫ130 mV holding potential rather than Ϫ150 mV. Changing the holding potential from Ϫ150 to Ϫ130 mV had no effect on parameters of SSSI in any of the constructs. The rate of recovery from SI was measured by depolarizing the membrane to 0 mV for 60 s to fully slow inactivate channels. The membrane potential was then stepped to an interpulse potential (Ϫ150 mV), 0.2-30 s in duration, before measuring the recovered current with a 0-mV test pulse. To measure the rate of SI onset, cells were held at Ϫ150 mV for 30 s, stepped to an prepulse potential (0 mV) for 0 -60 s, hyperpolarized to Ϫ150 mV for 20 ms to allow recovery of fast inactivated but not slow inactivated channels, and then available current was measured with a 0-mV test pulse.
Use-dependent inactivation was recorded as described previously (34). Briefly, the membrane was repetitively depolarized to 0 mV for 230 ms and hyperpolarized to Ϫ90 mV for 150 ms 500 times at a frequency of ϳ2.6 Hz. The frequency, which coincides with a 156-beat/min heart rate, was sufficient to generate a measurable reduction in current due to inactivation while remaining within the range of physiological depolarization and frequency values. The ratio of depolarization to repo-larization was chosen based on Bazett's formula to generate a healthy QT duration, calculated QTc ϭ 371 (36). No leak subtraction was used in use-dependent inactivation (UDI) recordings to ensure appropriate pulse frequency.
Data Analysis-Analysis of activation, FI, and SI data were completed using Fitmaster version 2x32 (HEKA Electronics) and Igor Pro version 5.01 (Wavemetrics, Lake Oswego, OR) run on an iMac. Conductance curves were computed from current/ voltage relationships using the equation, where G represents conductance, I max represents the peak test pulse current, V m is the test pulse voltage, and E rev is the measured reversal potential. Values were plotted as a function of test potential and then fitted with the Boltzmann equation, where z represents the apparent valence, e 0 is the elementary charge, V1 ⁄ 2 is the midpoint, T is the recording temperature in kelvin, and k is the Boltzmann constant. SSFI and SSSI data were fitted with a modified Boltzmann equation, where I Max and I Min represent the maximum and minimum values of the fit, respectively. All other values are identical to those in Equation 2. The time constants of the recovery and onset of FI were calculated from the single exponential equation, where Y 0 represents the asymptote of the fit, A is the relative component of the exponent, is the time constant, and x is time. The time constants of FI onset and recovery were plotted as a function of prepulse and interpulse voltage, respectively, and fitted with the Eyring model, where M represents the inverse of the maximum time constant, w is the reaction velocity, b is the barrier distance, V is the voltage, V P is the voltage at the peak of the fit, and T is the temperature. SI onset and recovery as well as UDI data were fitted with the double exponential equation, where Y 0 represents the asymptote of the fit, A 1 is the relative component of the first exponent, 1 is the slow time constant, A 2 is the relative component of the second exponent, 2 is the fast time constant, and x is time. Proton block data were plotted as a function of pH and were fitted with the Hill equation, where Y M and Y 0 represent the maximum and minimum values of the fit, respectively, X1 ⁄ 2 is the midpoint of the curve, X is the pH, and b is the rate. Statistical analysis was completed using Student's t test and an analysis of variance, where appropriate, using JMP (SAS Institute Inc., Cary, NC). All data are reported as mean Ϯ S.E., and statistical significance was taken at p Ͻ 0.05.

RESULTS
We sought to identify amino acid residues responsible for pH-dependent modulation of Na V 1.5 channel gating. Previous studies have isolated proton sensor residues within the p-loop of several Na V and K V channels (32,33,37,38). We therefore focused our efforts on the Na V 1.5 channel p-loop regions.
Activation and Proton Block-Reducing extracellular pH from pH 7.4 to pH 6.0 reduces single channel conductance and significantly depolarizes the midpoint (V1 ⁄ 2 ) and reduces the apparent valence (z) of activation of Na V 1.5 channels (32)(33)(34). In this study, the H880Q and C373F mutants displayed a significant reduction in proton block at pH 6.0 relative to WT channels ( Fig. 1). At pH 6.0, the maximum channel conductance was reduced by 38.7 Ϯ 2.0, 27.0 Ϯ 2.6, and 19.4 Ϯ 2.2% relative to pH 7.4 for WT, H880Q, and C373F channels, respectively ( Fig.  1, B, C, and D). To assess the profile of proton block for the three constructs, maximum conductance is plotted as a function of pH and fitted with a Hill curve (Equation 7) (Fig. 1E). The pK a values were not significantly different between the three con- Currents were normalized to the peak current amplitude recorded at pH 7.4 and plotted as a function of test potential. Proton block at pH 6.0 was significantly decreased in H880Q and C373F channels relative to WT: 27.0 Ϯ 2.6, 19.4 Ϯ 2.2, and 38.8 Ϯ 1.9%, respectively (p Ͻ 0.01, n ϭ 9 -13). E, conductance, normalized to the maximum conductance for each experiment (typically pH 8.0), is plotted as a function of pH and fitted with a Hill curve (Equation 7) for WT (squares), H880Q (circles), and C373F (triangles) channels. Asymptotes based on fit lines of the mean data are displayed for H880Q (dashed line, 6.3%), and C373F (dotted line, 15.2%) channels. Based on individual fits, the asymptotes of H880Q and C373F, but not WT, were significantly elevated from zero, 6.2 Ϯ 1.6, 8.5 Ϯ 1.7, and 0.2 Ϯ 2.9%, respectively (p Ͻ 0.05, n ϭ 4). Error bars, S.E. structs: WT, pK a ϭ 5.8 Ϯ 0.7; H880Q, pK a ϭ 5.7 Ϯ 0.04; C373F, pK a ϭ 5.9 Ϯ 0.1. The asymptotes of proton block, however, differed significantly from zero for H880Q and C373F (6.2 Ϯ 1.6 and 8.5 Ϯ 1.7%, respectively, p Ͻ 0.05, n ϭ 4) but not WT channels (0.2 Ϯ 2.9%).
We have previously shown that protons shift the voltage dependence of WT Na V 1.5 channel activation and SSFI and reduce the apparent valence of activation. Fig. 2 and Table 1 show that C373F and H880Q did not abolish this effect of protons. Surprisingly, the V1 ⁄ 2 of activation and SSFI in the H880Q mutant was significantly depolarized compared with C373F and WT channels, and the apparent valence of activation in both mutants was significantly greater than that in WT channels (Fig. 2). Window currents were estimated from the overlay of paired activation and SSFI curves. Peak window current occurred over a more depolarized voltage range in H880Q than C373F and WT channels: peak window current occurred at Ϫ51.3 Ϯ 1.8, Ϫ56.9 Ϯ 1.1, and Ϫ60.3 Ϯ 1.0 mV for H880Q, C373F, and WT, respectively (Fig. 2, C, D, and E). We also measured proton modulation of FI recovery (Ϫ150 through Ϫ70 mV), closed state FI (Ϫ90 through Ϫ40 mV), and open state FI (Ϫ40 through ϩ30 mV) in all three channel types (data not shown). In each case, the results reflected the previously described trends, where acidic pH causes a depolarizing shift in gating, comparable with previous findings (34,39), and H880Q gating is depolarized relative to WT and C373F channels. These data suggest that proton modulation of channel activation and slow inactivation are two independent processes.
Steady-state Slow Inactivation-Protons bind within the extracellular Na V channel pore (32,33). Given the mutual association of protons and SI at the outer turret, we hypothesized that residues within the pore mediate proton modulation of SI. We therefore measured the steady-state and kinetic parameters of SI in all three constructs with extracellular solution titrated to either pH 7.4 or pH 6.0. Fig. 3 displays SSSI recorded in WT, C373F, and H880Q channels at pH 7.4. Normalized current was plotted as a function of prepulse potential and fitted with a Boltzmann function (Equation 3). Both H880Q and C373F channels displayed an altered response to reduced extracellular pH compared with WT channels. Extracellular perfusion of pH 6.0 triggered a significant (albeit small) increase in the maximum probability of SSSI in H880Q and C373F channels (Fig. 3, C and D, and Table 2). pH 6.0 also reduced the apparent valence of SSSI in C373F but had no effect on the apparent valence of H880Q SSSI (Fig. 3, C and D, and Table 2). In contrast, reducing extracellular pH from pH 7.4 to pH 6.0 had no effect on the SSSI curve for WT channels (Fig. 3B). This correlates with previous recordings from Na V 1.5 channels expressed in oocytes (34). H880Q and C373F channels also displayed a reduced maximum probability of SSSI relative to WT channels, measured at pH 7.4 ( Table 2).

FIGURE 2. Proton modulation of activation, SSFI, and window current.
A, normalized conductance is plotted as a function of test potential and fitted with a Boltzmann function (Equation 2) for WT (squares), H880Q (circles), and C373F (triangles) channels. Reducing extracellular pH from pH 7.4 to pH 6.0 significantly depolarized the V1 ⁄2 and reduced the z of activation in WT and C373F channels but not H880Q, although all three constructs demonstrated a similar response to pH 6.0 (p Ͻ 0.05; see Table 1 for values). The V1 ⁄2 of the H880Q mutant was significantly depolarized compared with C373F and WT channels, Ϫ27.8 Ϯ 1.4, Ϫ34.3 Ϯ 1.3, and Ϫ33.6 Ϯ 1.1 mV, respectively (p Ͻ 0.05). The z values of activation of both H880Q and C373F were significantly larger than for WT channels (4.1 Ϯ 0.2 e, 3.9 Ϯ 0.2 e, and 3.1 Ϯ 0.1 e, respectively). B, proton modulation of SSFI. Normalized current is plotted as a function of prepulse potential and fitted with a Boltzmann function (Equation 3) for WT (squares), H880Q (circles), and C373F (triangles) channels. Reducing extracellular pH from pH 7.4 to pH 6.0 significantly depolarized the V1 ⁄2 of all three constructs but did not affect the z of SSFI. Again, the V1 ⁄2 of the H880Q mutant was significantly depolarized compared with C373F and WT channels at pH 7.4 but not pH 6.0 (Ϫ71.2 Ϯ 1.2 mV, Ϫ76.3 Ϯ 0.6 mV, and Ϫ77.3 Ϯ 0.4 mV, respectively). C-E, activation/SSFI overlays of Boltzmann fits displaying the similar trend of proton modulation of WT (C), H880Q (D), and C373F (E) channels recorded at pH 7.4 (solid lines) and pH 6.0 (dotted lines). The window current peaks, measured from the overlay paired activation and SSFI curves, reflected the voltage dependence of activation and SSFI. Window current peaks recorded at pH 7.4 are indicated by vertical dotted lines. H880Q was significantly right-shifted relative to C373F and WT channels (Ϫ51.3 Ϯ 1.8, Ϫ56.9 Ϯ 1.1, and Ϫ60.3 Ϯ 1.0 mV, respectively). The insets of A and B display the protocols used, respectively. Error bars, S.E.

JOURNAL OF BIOLOGICAL CHEMISTRY 4785
Slow Inactivation Recovery-We measured proton modulation of SI recovery at Ϫ90 mV because it correlated with the hyperpolarizing pulse of our UDI protocol (discussed below) and could therefore help explain changes in UDI between the three channel constructs. In each channel, SI recovery kinetics were biexponential. Protons accelerated SI recovery kinetics in WT but not H880Q and C373F channels (Fig. 4). In WT channels, pH 6.0 significantly increased slow and reduced the relative contribution of the slow component. Protons did not significantly affect fast but significantly increased the amplitude of the fast component (Fig. 4, A (inset) and B, and Table 3). Overall, the effects on the relative amplitudes of fast and slow decay were dominant to the changes in the fast and slow time constants, and extracellular acidosis accelerated SI recovery in WT channels (Fig. 4B). In H880Q channels, protons significantly increased fast as well as its relative component but did not affect slow (Fig. 4, A (inset) and C, and Table 3). The C373F mutation abolished proton modulation of SI recovery (Fig. 4, A (inset) and D, and Table 3). These data demonstrate that His-880 and Cys-373 both mediate modulation of SI recovery by protons and show that Cys-373 plays a greater role in mediating the response of SI to protons than His-880. Last, the apparent differences in recovery kinetics between constructs at pH 7.4 were not statistically significant ( Fig. 4A and Table 3).
Slow Inactivation Onset-We assessed proton modulation of SI onset at 0 mV to coincide with the depolarizing pulse of our UDI protocol and could therefore help explain changes in UDI between the three channel constructs. Protons displayed differential modulation of SI onset in WT, H880Q, and C373F channels ( Fig. 5 and Table 4). SI onset was slowed in WT channels at reduced extracellular pH (Fig. 5B). WT channels displayed significantly increased fast and slow at pH 6.0 relative to pH 7.4 ) in WT (squares), C373F (triangles), and H880Q (circles). H880Q and C373F channels displayed a reduced maximum probability of SSSI relative to WT channels (see Table 2 for values). B, reducing extracellular pH from pH 7.4 (solid lines) to pH 6.0 (dotted lines) had no effect on WT SSSI. pH 6.0 significantly increased SSSI maximum probability in H880Q (C) and C373F (D) channels from 46.0 Ϯ 1.8 to 49.5 Ϯ 1.6% and from 45.1 Ϯ 3.3 to 48.7 Ϯ 3.0%, respectively (n ϭ 5-11, p Ͻ 0.05). pH 6.0 also reduced the z of SSSI in C373F channels from Ϫ3.7 Ϯ 0.2 e to Ϫ2.5 Ϯ 0.2 e (n ϭ 6, p Ͻ 0.05). The inset in A depicts the protocol used. Error bars, S.E.    (Fig. 5, B and E, and Table 4). Extracellular perfusion of pH 6.0 did not affect the relative components of SI onset ( Fig. 5F and Table 4). SI onset was also slowed at pH 6.0 in the H880Q mutant, although the effect on fast was significantly reduced relative to WT (Fig. 5E). pH 6.0 increased fast in WT channels by 8.0 Ϯ 2.4 ms but increased fast by only 1.1 Ϯ 0.4 ms in H880Q channels. Conversely, C373F abolished proton modulation of fast and slow ; however, pH 6.0 significantly increased the relative component of slow (Fig. 5, D-F, and Table 4). Protons did not significantly affect the asymptote of SI onset in any of the constructs. These data demonstrate that His-880 and Cys-373 both mediate proton modulation of SI onset, and, as is the case for SI recovery, Cys-373 plays a greater role in mediating the response of SI to protons than His-880. The asymptote of SI onset at pH 7.4 was significantly different between the WT, H880Q, and C373F constructs (Fig. 5A and Table 4). Additionally, H880Q and C373F channels displayed a reduced slow component relative to WT channels ( Table 4). The differences in the SI onset asymptote between channel constructs are probably attributable to the observed reductions in the components of slow in H880Q and C373F.
Use-dependent Inactivation-Protons display profound isoform-specific modulation of UDI (39). UDI in Na V 1.4 is pH-insensitive, whereas in Na V 1.5, UDI is destabilized by protons. Inserting the p-loops of the skeletal muscle isoform, Na V 1.4, into the Na V 1.5 channel backbone abolishes proton modulation of UDI (40). We measured UDI using a pulse protocol designed to mimic a human ventricular action potential, as described previously (34). Cells were depolarized to 0 mV for 230 ms and then hyperpolarized to Ϫ90 mV for 150 ms to simulate an elevated heart rate of ϳ160 beats/min. This stimulation frequency is high enough to induce readily measurable levels of UDI but low enough to remain physiologically relevant. Peak currents from 500 consecutive pulses were recorded and normalized to the first pulse and then plotted as a function of time and fitted with a double exponential equation (Equation 6) as in Fig. 6.
In WT channels (Fig. 6A), we observed the previously described destabilization of UDI in response to reduced extracellular pH (34,39). WT channels displayed a significant increase in the asymptote of UDI at pH 6.0 relative to pH 7.4 ( Fig. 6A and Table 5). The C373F mutation removed the effects of protons on the asymptote of UDI but preserved proton modulation of Fast (Fig. 6C). C373F channels also displayed a significantly reduced Fast compared with WT (Table 5). H880Q completely abolished the effects of protons on UDI (Fig. 6, B, D, and E, and Table 5).
Last, both C373F and H880Q significantly reduced the relative components of Slow compared with WT under control conditions ( Table 5). The C373F mutant also displayed a significantly elevated UDI asymptote compared with WT and H880Q channels (Fig. 6 and Table 3). These data correlate well with the reduced maximum probability of SI observed in C373F and H880Q channels (Figs. 4 and 5). ) for WT (squares), H880Q (circles), and C373F (triangles) channels at pH 7.4. The asymptote of SI onset at pH 7.4 was significantly different between WT, H880Q, and C373F channels: 43.5 Ϯ 2.2, 48.6 Ϯ 1.2, and 57.9 Ϯ 1.5%, respectively (Table 4). B, pH 6.0 significantly increased ast and low but did not affect the relative component of either time constant in WT channels. C, proton modulation of SI recovery was preserved in the H880Q mutant, although the effect on fast was significantly reduced relative to WT (p Ͻ 0.05). D, C373F abolished proton modulation of fast and slow ; however, pH 6.0 significantly increased the relative component of slow .
The inset in D displays the pulse protocol used. Protons did not significantly affect the asymptote of SI onset in any of the constructs. E and F, bar graphs showing fast and slow (E) and relative components of fast and slow (F) for SI recovery at pH 7.4 (filled bars) and pH 6.0 (open bars). *, statistically significant at p Ͻ 0.05. Error bars, S.E.

DISCUSSION
Proton block and modulation of several Na V 1.5 gating parameters occurs with a pK a similar to a histidine (pK a ϳ6.0) (32,34). p-loop histidines are proton sensors in voltage-gated potassium channels, whereby replacement with a glutamine abolishes proton modulation of C-type inactivation (38,41). There are two p-loop histidines in Na V 1.5, His-880 and His-886. We created the Na V 1.5 mutants H880Q and H886Q, simulating constitutive deprotonation at those positions. The H886Q mutant, however, did not produce measurable currents. Cys-373 was shown previously to be involved in Na V 1.5 proton block and seemed a likely candidate to modulate slow inactivation (32). We thus recorded ionic currents from three Na V 1.5 channel constructs: WT and two p-loop mutants, C373F and H880Q. We found that H880Q shifted the voltage dependence of activation relative to WT channels, reduced the pH-dependent modulation of slow inactivation kinetics, and abolished the pH dependence of UDI. C373F did not affect activation but removed the pH-dependent modulation of SI kinetics and reduced the pH dependence of UDI. Last, both mutations reduced proton block and destabilized SI relative to WT channels.
Molecular Determinants of Proton Block-Unlike Na V 1.5, the skeletal muscle isoform, Na V 1.4, is not fully blocked by protons (32,33). Na V 1.4 channels display a pH-insensitive current that is roughly ϳ14% of maximum conductance (33). In Na V 1.4, replacement of Tyr-402 with a cysteine (the homologous residue in Na V 1.5) abolishes the proton-insensitive current (32). The reverse mutation in Na V 1.5, C373Y, imparts a proton-insensitive current similar to that seen in Na V 1.4 (32). Replacement of Tyr-402 with a phenylalanine or a serine, the equivalent residues in Na V 1.2 and Na V 1.8, respectively, does not alter the pH-insensitive current, suggesting that Na V 1.2 and Na V 1.8 would have a proton block asymptote similar to that of Na V 1.4 (32,39). Kahn et al. (32) suggested that protonation of Cys-373 along with the outer pore carboxylates Glu-375, Glu-901, Asp-1423, and Asp-1714 imparts proton block in Na V 1.5 and that the absence of this pore cysteine in other Na V channels results in a pH-insensitive current.
Here we report that two novel mutations, C373F and H880Q, alter Na V 1.5 channel proton block. Both mutations significantly reduced the degree of proton block observed at pH 6.0 relative to WT channels and displayed asymptotes of proton block that represented a pH-insensitive component of current  6). Error bars are omitted for image clarity. A, pH 6.0 (gray symbols) significantly increased the Fast from 4.0 Ϯ 0.5 to 7.2 Ϯ 0.7 s and the asymptote of UDI from 60.6 Ϯ 1.1 to 65.3 Ϯ 1.4% in WT channels. B, the C373F mutation removed proton modulation of the UDI asymptote but not Fast . pH 6.0 significantly increased Fast in C373F channels from 2.5 Ϯ 0.2 to 5.0 Ϯ 0.6 s. Additionally, Fast in C373F channels was significantly reduced compared with WT channels (Table 5). C, UDI in H880Q channels was not modulated by protons. D and E, bar graphs depicting Fast (D) and asymptote (E) of UDI of WT, C373F, and H880Q channels at pH 7.4 (filled bars) and pH 6.0 (open bars). The inset of B displays the pulse protocol used. *, statistically significant at p Ͻ 0.05. Error bars, S.E.

TABLE 5
Use-dependent inactivation (Fig. 1). C373F proton block data fitted with a Hill curve (Equation 7) displayed a pH-insensitive conductance that was roughly 9% of maximum conductance (Fig. 1E), consistent with previous results from the C373Y Na V 1.5 mutant (32). H880Q also displayed a proton block asymptote that was significantly different from zero at ϳ6% of maximum conductance (Fig. 1E). His-880 has not been implicated previously in Na V channel proton block, and it is interesting that H880Q proton block was intermediate between that observed in C373F and WT channels. Kahn et al. (33) hypothesized that protonation at the p-loop increases the electrostatic potential within the Na V channel outer vestibule, thereby repelling Na ϩ ions. Our results further support that hypothesis. Cys-373 is adjacent to the selectivity filter, where the electrostatic potential is the most negative. His-880, however, is predicted to be at the top of the Na V channel P1 helix of DII (Fig. 7). Based on their relative locations, it is understandable that the effect of the H880Q mutation would be less dramatic than that observed by the C373F mutation.
Pore Mutants Destabilize Slow Inactivation-Na V 1.2 channels inactivate more completely than Na V 1.5; during prolonged depolarization, more than 65% of Na V 1.2 channels slow inactivate compared with ϳ50% in Na V 1.5 (8,14,34,42). Because the homologous position to Na V 1.5 Cys-373 is Phe-385 in Na V 1.2 channels, and because SI is known to be sensitive to mutations within the external pore, we predicted that the C373F mutant would confer a stabilization of slow inactivated states in Nav1.5 (Fig. 7) (8,14,32). Because His-880 is highly conserved (Fig. 7), we also predicted that the H880Q mutation would disrupt Na V 1.5 SI. Interestingly, both H880Q and C373F destabilized the slow inactivated state. The maximum probability of SSSI was reduced by ϳ10% in H880Q and C373F channels relative to WT (Table 2). Additionally, compared with WT channels, the asymptote of SI onset measured at 0 mV was increased in H880Q and C373F channels by ϳ10 and ϳ19%, respectively ( Table 4). Destabilization of SI by C373F was further implicated by an observed increase in the asymptote of UDI, ϳ12% ( Table  5). The asymptote of H880Q UDI displayed an elevated trend (ϳ4%), but the difference from WT was not statistically significant (Table 5).
Although alterations to SI by H880Q were predicted because His-880 is highly conserved throughout mammalian Na V isoforms, destabilized SI in C373F channels was surprising. As mentioned previously, Cys-373 in Na V 1.5 is homologous to Phe-385 in Na V 1.2 (Fig. 7), and, because Na V 1.2 has more complete SI than Na V 1.5, C373F was expected to enhance SI (8,14,34,42). In contrast to expectations, C373F reduced the maximum probability of SI. When Ile-891, positioned in the Na V 1.5 p-loop (Fig. 7), is replaced with a valine, the homologous residue in Na V 1.4, the mutant channel displays Na V 1.4-like SI (14). The reverse mutation in Na V 1.4 channels confers Na V 1.5-like SI (14). Na V 1.2 also has a valine at this position (Val-935; Fig. 7). Our data suggest that the differences in maximum probability of SI between Na V 1.5 and Na V 1.2 channels may not be isolated to a single residue as is the case between Na V 1.5 and Na V 1.4.
Proton Sensors and Slow Inactivation-We measured UDI using a pulse protocol designed to mimic a human ventricular action potential cycling at 2.6 Hz or ϳ160 beats/min, as described previously (34). Given the length of the depolarizing and hyperpolarizing pulses (230 and 150 ms, respectively), this protocol is an effective tool to assess the dynamic equilibrium of slow inactivated states, which have onset and recovery time constants in the range of seconds to tens of seconds (Tables 3  and 4). Fast inactivation, with onset and recovery time constants less than 30 ms, has a negligible contribution to the decay of current during this experimental protocol (34). Our data demonstrate that p-loop residues mediate proton modulation of UDI of Na V 1.5 channels.
Both H880Q and C373F abolished proton modulation of UDI. Protons increase, reduce, or have no effect on the asymptote of UDI in Na V 1.5, Na V 1.2, and Na V 1.4 channels, respectively (34,39). Inserting the p-loops of Na V 1.4 into the Na V 1.5 channel backbone imparts Na V 1.4 channel-like proton modulation of UDI (40). The p-loops between isoforms are highly conserved, and Cys-373 is one of only a few positions that differ between the three channels. As mentioned previously, Cys-373 is responsible for isoform-specific proton block between cardiac and skeletal channels (32). The homologous residues in Na V 1.2, Na V 1.4, and Na V 1.8 channels (phenylalanine, tyrosine, and serine, respectively) impart a proton-insensitive current in Na V 1.4, whereas cysteine abolishes it (32). We therefore postulate that protonation of Cys-373 modulates UDI in Na V 1.5 and contributes to the differences in UDI modulation between cardiac and skeletal channel isoforms. Further, our results with H880Q and C373F suggest that accumulation of positive charge around the Na V 1.5 channel pore mediates destabilization of UDI and that Cys-373 protonation represents a "tipping point" of positive charge that induces the reduction in Na V 1.5 UDI. These hypotheses require the assumption that SI disruption under control conditions in H880Q and C373F relative to WT is allosteric and not electrostatic.
These data paired well with our data on the kinetics of SI. C373F was more effective than H880Q at abolishing the pH-de- FIGURE 7. A, predicted Na V channel secondary structure depicting the relative locations of the C373F (ૺ) and H880Q (ࡗ) mutations. B, sequence alignment of the p-loop segments in bacterial NaChBac, Na V Ab, and DI and DII of the mammalian Na V 1.2 (accession number Q99250.3) and Na V 1.5 (accession number Q14524.2) channels (1). Selectivity residues of NaChBac and Na V Ab along with the homologous residues in Na V 1.2 and Na V 1.5 channels are highlighted in gray. Residues that differ between Na V 1.2 and Na V 1.5 are encapsulated. P1 and P2 represent the two pore helices isolated in the crystal structure of the Na V Ab channel (17,43).
Proton Sensors in Na V 1.5 FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 pendent modulation of SI kinetics (Figs. 4 and 5 and Tables 3  and 4). SSSI in C373F also displayed a greater response to protons than H880Q, whereas WT channels showed no change ( Fig. 3 and Table 2). Overall, H880Q showed intermediate responses to protons compared with WT and C373F channels (Tables 2-4). Like their role in proton block, the relative positions of His-880 and Cys-373 may underlie the differential contribution to SI proton modulation. Because slow inactivation is thought to involve a structural rearrangement at or around the selectivity filter, protonation of Cys-373 might be predicted to have a greater effect on SI than His-880, consistent with our results.
Conclusion-Our results demonstrate that p-loop residues His-880 and Cys-373 mediate Na V 1.5 sensitivity to protons. These results also identified Cys-373 as a residue responsible for isoform-specific proton modulation of UDI. It seems likely that proton block and SI modulation occur in tandem, given the strong overlap in results between the two processes. Additionally, these results suggest that His-880 and Cys-373 contribute to the stability of SI in Na V 1.5 channels.